RESEARCH ARTICLE
The Oral Microbiome of Denture Wearers Is Influenced by Levels of Natural Dentition Lindsay E. O’Donnell1, Douglas Robertson1, Christopher J. Nile1, Laura J. Cross1, Marcello Riggio1, Andrea Sherriff1, David Bradshaw2, Margaret Lambert2, Jennifer Malcolm1, Mark J. Buijs3, Egija Zaura3, Wim Crielaard3, Bernd W. Brandt3, Gordon Ramage1* 1 Glasgow Dental School, School of Medicine, College of Medicine, Veterinary and Life Sciences, University of Glasgow, 378 Sauchiehall Street, Glasgow, G2 3JZ, United Kingdom, 2 GlaxoSmithKline, St Georges Avenue, Weybridge, Surrey, United Kingdom, 3 Department of Preventive Dentistry, Academic Centre for Dentistry Amsterdam, University of Amsterdam and VU University Amsterdam, Amsterdam, the Netherlands * [email protected]
Abstract OPEN ACCESS Citation: O’Donnell LE, Robertson D, Nile CJ, Cross LJ, Riggio M, Sherriff A, et al. (2015) The Oral Microbiome of Denture Wearers Is Influenced by Levels of Natural Dentition. PLoS ONE 10(9): e0137717. doi:10.1371/journal.pone.0137717 Editor: Zezhang Wen, LSU Health Sciences Center School of Dentistry, UNITED STATES
Objectives The composition of dental plaque has been well defined, whereas currently there is limited understanding of the composition of denture plaque and how it directly influences denture related stomatitis (DS). The aims of this study were to compare the microbiomes of denture wearers, and to understand the implications of these towards inter-kingdom and host-pathogen interactions within the oral cavity.
Received: July 27, 2015
Methods
Accepted: August 21, 2015
Swab samples were obtained from 123 participants wearing either a complete or partial denture; the bacterial composition of each sample was determined using bar-coded illumina MiSeq sequencing of the bacterial hypervariable V4 region of 16S rDNA. Sequencing data processing was undertaken using QIIME, clustered in Operational Taxonomic Units (OTUs) and assigned to taxonomy. The dentures were sonicated to remove the microbial flora residing on the prosthesis, sonicate was then cultured using diagnostic colorex Candida media. Samples of unstimulated saliva were obtained and antimicrobial peptides (AMP) levels were measured by ELISA.
Published: September 14, 2015 Copyright: © 2015 O’Donnell et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Funding: This study was funded by the BBSRC CASE studentship (BB/K501013/1) awarded to Prof Gordon Ramage and Dr Douglas Robertson. GlaxoSmithKline (GSK) also supported this project. GSK were involved in the initial discussions at the conception of the study. GlaxoSmithKline provided support in the form of salaries for authors DB and ML, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific
Results We have shown that dental and denture plaques are significantly distinct both in composition and diversity and that the oral microbiome composition of a denture wearer is variable and is influenced by the location within the mouth. Dentures and mucosa were predominantly made up of Bacilli and Actinobacteria. Moreover, the presence of natural teeth has a significant impact on the overall microbial composition, when compared to the fully edentulous. Furthermore, increasing levels of Candida spp. positively correlate with Lactobacillus spp. AMPs were quantified, though showed no specific correlations.
PLOS ONE | DOI:10.1371/journal.pone.0137717 September 14, 2015
1 / 22
Oral Microbiome Analysis of Denture Wearers
roles of these authors are articulated in the "author contributions" section. Competing Interests: GlaxoSmithKline (GSK) supported this project and were involved in the initial discussions at the conception of the study. David Bradshaw and Margaret Lambert are employed by GlaxoSmithKline and Gordon Ramage acted as a key opinion leader for GSK. There are no patents, products in development or marketed products to declare. This does not alter the authors' adherence to all the PLOS ONE policies on sharing data and materials, as detailed online in the guide for authors.
Conclusions This is the first study to provide a detailed understanding of the oral microbiome of denture wearers and has provided evidence that DS development is more complex than simply a candidal infection. Both fungal and bacterial kingdoms clearly play a role in defining the progression of DS, though we were unable to show a defined role for AMPs.
Introduction Healthcare improvements in the last century have led to an increasingly elderly population. Worldwide, 810 million people are aged 60 years or over, which is predicted to increase to at least two billion by 2050 (22% of the entire global population) [1]. In the EU alone the proportion of the population who are 65 years and older is predicted to reach 53% by the year 2025 [2]. This demographic change will result in significant challenges for oral healthcare delivery, to an increasingly aged population with declining oral health. As the population ages then oral diseases become more relevant with respect to their local and systemic impact, which can have profound implications for healthcare provision [3, 4]. The oral cavity is a complex environment that is continually exposed to numerous opportunistic microbial pathogens. These are kept in check by a robust arsenal of immune factors that maintain a healthy oral environment and prevent the development of disease. This arena has gradually become a key area of biomedical research, which has led to a greater understanding of the causes, pathogenesis and host response against oral disease, with the majority of research focussing on diseases affecting dentate individuals, such as gingivitis, periodontitis and caries. Conversely, there is very limited research regarding denture related disease. Despite major improvements in oral health worldwide, recent estimates report that the rate of edentulouness still varies from 7 to 69% of the worlds adult population [5, 6], and in the US and UK populations around one fifth wear some form of removable denture [7, 8]. This continued high prevalence should convince researchers that, there is a requirement to develop an understanding of the implications of dentures on oral and systemic health. The primary condition denture wearers suffer from is denture stomatitis (DS). This refers to inflammation of the oral mucosa and pathological changes associated with the wearing of dentures [9]. DS can be classified according to the severity of inflammation using a scale first described by Newton [10, 11]. The aetiology of DS is related to a variety of factors including poorly fitting dentures causing trauma and biological factors such as poor salivary flow, smoking or antibiotic treatment as well as microbial infection [12]. However, Candida albicans is generally attributed as being the main causative agent in DS affecting approximately 30–70% of denture wearers [9]. Candida spp. colonise the denture surface, forming co-aggregates with bacteria to build complex microbial communities known as biofilms. The majority of literature in this area focuses solely on Candida as the primary cause of infection, however, there is increasing evidence to suggest that this is very much a polymicrobial disease in which bacterial and fungal interactions play a role in disease pathogenesis [13, 14]. Several studies have isolated bacteria directly from the surface of dentures using standard microbial culturing techniques, primarily streptococci and staphylococci species [15–18]. However, these culture based methods are largely inadequate and unlikely to give a comprehensive representation of the polymicrobial population, which can contain up to 1011 microbes per milligram of denture plaque [19]. The advent of high throughput sequencing however has revolutionised our understanding of microbial ecosystems, and thus using this superior method we can for the first time gain an insight into the oral microbiome of denture wearers.
PLOS ONE | DOI:10.1371/journal.pone.0137717 September 14, 2015
2 / 22
Oral Microbiome Analysis of Denture Wearers
In addition to an incomplete understanding of the composition of denture plaque, we also have limited knowledge and understanding of the local host immune response. It has been established that the immune response is gradually impaired with increasing age [20, 21], but with the addition of loss of natural teeth an even greater rapid decline in host protective responses in the oral cavity is reported [22]. Antimicrobial peptides (AMP) including cathelicidin LL-37, histatins and defensins, which exhibit antimicrobial and immunoregulatory properties and protect mucosal surfaces against pathogens, have been found to be present in saliva. However, no studies have yet investigated salivary AMPs in denture wearers and the potential role they may play in the protection against denture plaque. Thus, a more in depth investigation is required into the host-microbiome relationship in the denture wearing individuals, which may help towards understanding the pathogenesis of DS. The aim of this study was therefore to carry out a detailed site specific analysis of the bacterial composition of the oral microbiome of denture wearers using high throughput 16S rRNA gene sequencing technology. Furthermore, to assess changes in the diversity and composition of the microbiome against candidal load to aid understanding of the potential bacterial-fungal interactions that may be important in DS. In addition, the relationship between host antimicrobial peptides and denture plaque composition was also investigated. This is the first study to provide such a detailed microbial analyses of denture biofilms.
Materials and Methods Study participants 123 denture wearing patients attending the University of Glasgow Dental School and Hospital were enrolled in the study. Convenience sampling was used based on the patient availability on recruitment days. Patients were recruited by a designated research nurse or a PhD student. Written informed consent was obtained from all participants. Ethical approval for the study was granted by the West of Scotland Research Ethics Service (12/WS/0121). All patients wore full or partial removable dentures. A team of qualified dental clinician were responsible for the collection of samples and the recording of clinical features. Data on patient age, gender, smoking status, routine oral hygiene regimens and any history of recent antimicrobial medication were recorded on a patient questionnaire. There was no age related exclusion criteria for this study. Newton’s classification method for DS was used to score the appearance of the patient’s palatal mucosa [11]. The following scores were applied; 0 = healthy mucosa, 1 = pin-point hyperaemic lesions (localized erythema), 2 = diffuse erythema (generalized simple inflammation), and 3 = hyperplastic granular surface (inflammatory papillary hyperplasia). For standardisation, all clinicians received training to calibrate scoring the extent of erythema. Patients were excluded from the study if they were pregnant, had previous radiotherapy for the treatment of head and neck malignancy, had been receiving antimicrobial/antifungal treatment, using prescription mouthwashes or had received immunosuppressant therapy within six months previous to sampling.
Clinical sample collection Clinical samples were collected as shown in Fig 1. Ethylene oxide sterilised swabs (Fisher Scientific, Loughborough, UK) were used to collect samples from the denture surface in contact with the palatal mucosa and the palatal mucosal surface covered by the denture. If any natural were teeth present the clinician took a supragingival plaque sample from a single tooth using a sterile dental probe, which was immediately placed into a 2 ml collection tube (Fisher Scientific) containing RNAlater1 (QIAgen, Manchester, UK). Dentures were removed from the patients’ mouth and placed in sterile bags (Fisher Scientific) filled with 50 ml PBS (Sigma-Aldrich,
PLOS ONE | DOI:10.1371/journal.pone.0137717 September 14, 2015
3 / 22
Oral Microbiome Analysis of Denture Wearers
Fig 1. Flow chart demonstrating the process of clinical sample collection and processing in the study. doi:10.1371/journal.pone.0137717.g001 PLOS ONE | DOI:10.1371/journal.pone.0137717 September 14, 2015
4 / 22
Oral Microbiome Analysis of Denture Wearers
Dorset, UK), then placed in a sonic bath (Ultrawave, Cardiff, UK) for 5 min to remove adherent denture plaque. The denture sonicate was then transferred to a 50 ml tube. Denture sonicate was centrifuged for 10 min at 3700 x g, and the plaque pellet re-suspended in 2 ml of RNAlater (QIAgen), as previously described [23]. Swab tips were removed and stored in RNAlater. Denture plaque, dental plaque and swab samples were all stored at -80°C. In total samples from 131 patients were collected, which included 131 denture swabs, 131 mucosal swabs and 79 dental plaque samples. During DNA extraction process not all samples had sufficient DNA to use for sequencing, therefore DNA from 108 denture samples, 87 mucosal samples and 63 dental samples remained for sequencing, collectively all these samples originated from 123 patients. Samples were further separated into distinct groups for direct comparison and analysis: health and DS groups, dentate and edentate groups and complete denture and partial denture groups.
Collection of saliva Samples of whole unstimulated saliva were obtained by expectorating into a SaivaBio collection tube (Salimetrics, Suffolk, UK). Patients were given a maximum of 5 min to provide a sample. Saliva was clarified by centrifugation at 10,000 g for 10 min and then stored at -80°C.
Candida isolation One ml of denture sonicate was used to prepare ten-fold serial dilutions ranging from 100– 10−3. The dilutions were then used to perform colony forming unit (CFU) counts using Colorex Candida plates on which C. albicans colonies appeared green, C. glabrata were pink and C. tropicalis were blue. (E&O Laboratories, Bonnybridge, UK). One hundred μl of each serial dilution was spread across each plate, and then incubated at 30°C for 48 hrs. CFUs forming on plates were then counted and the average number of Candida cells colonising each denture was calculated.
DNA isolation Dental plaque samples were centrifuged for 15 min @ 13000 x g the supernatant was removed and the sample resuspended in 150 μl TE buffer. Swab samples were sonicated for 30 sec, and then the sonicated fluid was transferred into a deep well plate and centrifuged for 15 min at 13000 x g. The precipitate was then resuspended in 150 μl TE buffer. All samples were then transferred to a plate with each well containing 0.25 ml of lysis buffer (AGOWA mag Mini DNA Isolation Kit, AGOWA, Berlin, Germany), 0.3 g zirconium beads (diameter, 0.1 mm; Biospec Products, Bartlesville, OK, USA) and 0.2 ml phenol. The samples were homogenized with a Mini-beadbeater (Biospec Products) for 2 min. DNA was extracted with the AGOWA mag Mini DNA Isolation Kit.
Quantitative PCR Real time qPCR was performed to determine the concentration of bacterial DNA per sample as the original DNA extracted may contain DNA from other microbes such as fungi. Primers and a probe for the 16S rRNA gene were used, (F: TCCTACGGGAGGCAGCAGT, R: GGACTAC CAGGGTATCTAATCCTGTT Probe: 6FAM-CGTATTACCGCGGCTGCTGGCAC-BBQ). As an internal control for PCR inhibition a qPCR of PhHV (Phocid herpesvirus type 1 gB gene) was also performed as described by Watzinger et al (2004) [24]. The total reaction volume was 20 μl, including 3 μl of DNA. Reactions contained a 26PCR Probe Master Mix (Roche), for 16S rRNA, 7.5 pmol primers and 3.8 pmol probe. For PhHV 1.8 pmol primers
PLOS ONE | DOI:10.1371/journal.pone.0137717 September 14, 2015
5 / 22
Oral Microbiome Analysis of Denture Wearers
and 0.4 pmol probe of each primer was used. qPCR was carried out using the Light cycler LC480-II (Roche Diagnostics, Switzerland) under the following conditions: an activation step of 10 min at 95°C, followed by 50 cycles consisting of a denaturation step at 95°C for 30 sec, an annealing step at 60°C for 30 sec, and an extension step at 72°C for 30 sec. Bacterial 16S rDNA concentrations (CFU/ml) were determined from standard curves of E. coli K12 cultures.
PCR amplification and Illumina sequencing Amplicon libraries of the V4 hypervariable region of the 16S rRNA gene were generated for each of the individual samples. PCR was performed using the forward primer 515F (GTGCCA GCMGCCGCGGTAA) and the reverse primer 806R (GGACTACHVGGGTWTCTAAT) [25]. The primers included Illumina adapters and a unique 8-nt sample index sequence key [26]. The amplification mix contained 2 units of Phusion HotStart II High fidelity polymerase (Thermoscientific), 1 unit Buffer Phusion HS II [5x], including 1.5 mM MgCl2 (Thermoscientific), 0.2 mM dNTP (Thermoscientific, Germany) and 1 μM of each primer. To each reaction 1 ng of DNA template was added. After denaturation (98°C; 30 sec), 35 cycles of denaturation (98°C; 10 sec), annealing (55°C; 30 sec), and extension (72°C; 30 sec) were performed. Individual amplicon libraries were analyzed for DNA content with the fluorescent Quant-iT™ PicoGreen1 dsDNA Assay Kit (Invitrogen). The libraries were pooled in equimolar amounts. The amplicons were purified by means of the IllustraTM GFXTM PCR DNA and Gel Band Purification Kit (GE Healthcare, Eindhoven, the Netherlands). The quality and the size of the amplicons were analyzed on the Agilent 2100 (Santa Clara, CA, USA). The amplicon was sequenced in paired end mode on a MiSeq sequencing system (Illumina, Eindhoven, the Netherlands) with the v2 kit (Illumina) [26, 27].
Sequencing data analysis Reads were first quality filtered using Trimmomatic v0.32, [28]. Next, the reads were merged using fastq-join implemented in QIIME v.1.8.0 [29]. Sequences were clustered into operational taxonomic units (OTUs) using USEARCH v7.01090 [30], after quality filtered with usearch (maxee 0.5).The representative sequence of each cluster was assigned a taxonomy using the RDP classifier [31]. The Ribosomal Database Project: improved alignments and new tools for rRNA analysis. Nucl Acids Res 37: D141–145. doi: 10.1093/nar/gkn879) (QIIME v.1.8.0) (Greengenes v13.8 97_otus set) with a minimum confidence of 0.8.
ELISAs A range of antimicrobial peptides which are commonly associated with the oral cavity were selected for assessment in our patient’s saliva [32–34]. The following AMPs were assessed by ELISA, LL-37, Calprotectin, Lactoferrin, HNP1-3 (Hycult biotech, The Netherlands), Histatin 5 (Stratech scientific, Suffolk, UK) and Beta defensin 1 (BD1) (Peprotech, London, UK) as per manufacturer’s instructions. Clarified saliva samples were diluted 1:5 in assay buffer (PBS, 0.5% BSA, 0.1% Tween20). Results were calculated using a 4-parameter curve fit, quantifying colorimetric changes at 630 nm (BMG-Labtech, Ortenberg, Germany).
Statistical analyses The data set was randomly sub-sampled to 770 reads per sample (minimum number of reads per sample was 776) to include the maximum number of samples for analysis. OTU datasets were reduced by log2 transformation so as to carry out principal component analysis (PCA)
PLOS ONE | DOI:10.1371/journal.pone.0137717 September 14, 2015
6 / 22
Oral Microbiome Analysis of Denture Wearers
and diversity statistics (Shannon diversity index and Dominance index); the analysis was carried out using PAST software [35]. A one-way ANOVA test was applied to compare diversity statistics at oral microbiome sites using GraphPad Prism software (version 4; La Jolla, CA, USA). PCA was used to reduce the dimensionality of the OTU dataset. 258 OTU’s were entered into the PCA. A scree plot was used to determine how many components emerged. Factor loadings above 0.15 on a component were considered to have a strong association with that component and were deemed to be the most informative in describing the microbiome components. To determine if distinct clusters formed for each group on the PCA plots, new variables were created for each principle component by using the factor loadings as regression coefficients, producing a score for each sample. These scores were then used as outcome variables to compare between groups (using t-tests where appropriate- dentate/edentate and complete/partial dentures groups). The contribution of each bacterial class was calculated in terms of proportion to the overall sample, percentages were log transformed and a t-test was used to compare health and DS groups. Diversity statistics were compared via a t-test with GraphPad Prism v5. Spearman’s rank correlation was used to assess correlations between the abundance of individual bacterial classes or genera with the proportion of Candida found on dentures (CFU counts), using SPSS version 20. Salivary concentrations of AMPs were compared between healthy and DS groups, dentate/ edentate groups and complete/partial groups (S1 Table). Data were log transformed and analysed using a t-test with prism. Furthermore Spearman’s rank correlation was used to assess correlations between the abundance of individual bacterial classes or genera with the concentrations of salivary AMPs found on dentures. To visualize the relationships and associations of the microbiomes with environmental variables canonical correspondence analysis (CCA) was applied. This form of analysis, carried out using PAST, allows the visualisation of OTU distribution and sample group distribution in relation to a number of environmental variables. Environmental variables included were Candida CFU counts and salivary concentration of a number of AMPs. The significance of each of the CCA axes was calculated by permuting the data 999 times. Candida CFU counts were compared between healthy and DS groups, dentate/edentate groups and complete/partial groups. Data were log transformed and analysed using a t-test with GraphPad Prism v5.
Study design The study was designed as a pilot study and was initially only powered to detect a biologically meaningful association between diseased and healthy mouths and microbiome composition, and therefore was not originally powered to detect differences between additional variables including, denture type, dentate status, Candida levels and salivary AMP levels. Thus, non-significant results between these variables are not necessarily absence of effect, but a result of not achieving the full sample size required. S2 Table provides a list of the statistics carried for all the analysis. Dentate and edentate groups were compared separately from complete and partial denture groups, as having a complete denture only applied to the upper denture from which the sample was taken. The patient however may have had natural teeth remaining on the mandible and therefore could not be classed as edentulous. The inter-correlation between complete and edentulous or partial and dentate groups are shown in Table 1. Thus given the variation in group numbers, analysis was carried out on both datasets.
PLOS ONE | DOI:10.1371/journal.pone.0137717 September 14, 2015
7 / 22
Oral Microbiome Analysis of Denture Wearers
Table 1. Patient demographics.
Patients n = 123
Gender n (%)
Mean age
Type of Denture n (%)
Dentate status
Disease status n (%)
Mean age of denture
(Male) (Female)
(years)
(Complete upper) (Partial upper)
(Edentulous) (Dentate)
(Healthy) (Disease)
(years)
43 (35)
70.7 ± 11.5 (min: 33, max: 95)
4.6 ± 5.3 (min:0.2, max: 40)
80 (65)
82 (67)
56 (46)
78 (63)
41 (33)
67 (54)
45 (37)
doi:10.1371/journal.pone.0137717.t001
Results Patient demographics Of the 123 patients that participated in this study, 43 were male and 80 were female (Table 1). The mean (SD) patient age was 70.7 (11.5) years (min: 33, max: 95) with an average (SD) denture age of 4.6 (5.3) years (min: 0.2 max: 40). Clinical diagnoses indicated that 63% of participants had a healthy oral mucosa and the remaining 37% had an inflamed mucosa of varying severity, and were suffering from DS. The majority of patients (67%) wore complete upper dentures, while the remainder (33%) had a partial upper denture with 1 natural teeth remaining.
Illumina sequencing output Across all of the samples 632 OTUs were identified, with an average of 94 OTUs per sample (SD 39; min 25, max 254), 536 of which contained a minimum of 5 reads. The data was subsampled to 770 reads per sample in order to avoid bias of variable sample size. After sub-sampling 502 OTUs remained with an average of 46 OTUs per sample (SD 18; min 11, max 121). The samples were categorised into five main phyla which represented 99.6% of the reads: Firmicutes (40.6%), Actinobacteria (23%), Bacteriodetes (22.2%), Proteobacteria (9.8%) and Fusobacteria (4%).
Oral microbiome by sample site Samples were categorised into groups according to sample site: denture (n = 108), mucosal (n = 87) and dental (n = 63), with 337, 414 and 306 OTUs identified, respectively. On average, denture samples had 42 OTUs (SD 14; min 13, max 87), mucosal 49 OTUs (SD 20; min 11, max 121) and dental 51 OTUs (SD 37; min 16, max 306). Interestingly, denture plaque samples were found to have significantly less OTUs compared to dental, (p
The Oral Microbiome of Denture Wearers Is Influenced by Levels of Natural Dentition Lindsay E. O’Donnell1, Douglas Robertson1, Christopher J. Nile1, Laura J. Cross1, Marcello Riggio1, Andrea Sherriff1, David Bradshaw2, Margaret Lambert2, Jennifer Malcolm1, Mark J. Buijs3, Egija Zaura3, Wim Crielaard3, Bernd W. Brandt3, Gordon Ramage1* 1 Glasgow Dental School, School of Medicine, College of Medicine, Veterinary and Life Sciences, University of Glasgow, 378 Sauchiehall Street, Glasgow, G2 3JZ, United Kingdom, 2 GlaxoSmithKline, St Georges Avenue, Weybridge, Surrey, United Kingdom, 3 Department of Preventive Dentistry, Academic Centre for Dentistry Amsterdam, University of Amsterdam and VU University Amsterdam, Amsterdam, the Netherlands * [email protected]
Abstract OPEN ACCESS Citation: O’Donnell LE, Robertson D, Nile CJ, Cross LJ, Riggio M, Sherriff A, et al. (2015) The Oral Microbiome of Denture Wearers Is Influenced by Levels of Natural Dentition. PLoS ONE 10(9): e0137717. doi:10.1371/journal.pone.0137717 Editor: Zezhang Wen, LSU Health Sciences Center School of Dentistry, UNITED STATES
Objectives The composition of dental plaque has been well defined, whereas currently there is limited understanding of the composition of denture plaque and how it directly influences denture related stomatitis (DS). The aims of this study were to compare the microbiomes of denture wearers, and to understand the implications of these towards inter-kingdom and host-pathogen interactions within the oral cavity.
Received: July 27, 2015
Methods
Accepted: August 21, 2015
Swab samples were obtained from 123 participants wearing either a complete or partial denture; the bacterial composition of each sample was determined using bar-coded illumina MiSeq sequencing of the bacterial hypervariable V4 region of 16S rDNA. Sequencing data processing was undertaken using QIIME, clustered in Operational Taxonomic Units (OTUs) and assigned to taxonomy. The dentures were sonicated to remove the microbial flora residing on the prosthesis, sonicate was then cultured using diagnostic colorex Candida media. Samples of unstimulated saliva were obtained and antimicrobial peptides (AMP) levels were measured by ELISA.
Published: September 14, 2015 Copyright: © 2015 O’Donnell et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Funding: This study was funded by the BBSRC CASE studentship (BB/K501013/1) awarded to Prof Gordon Ramage and Dr Douglas Robertson. GlaxoSmithKline (GSK) also supported this project. GSK were involved in the initial discussions at the conception of the study. GlaxoSmithKline provided support in the form of salaries for authors DB and ML, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific
Results We have shown that dental and denture plaques are significantly distinct both in composition and diversity and that the oral microbiome composition of a denture wearer is variable and is influenced by the location within the mouth. Dentures and mucosa were predominantly made up of Bacilli and Actinobacteria. Moreover, the presence of natural teeth has a significant impact on the overall microbial composition, when compared to the fully edentulous. Furthermore, increasing levels of Candida spp. positively correlate with Lactobacillus spp. AMPs were quantified, though showed no specific correlations.
PLOS ONE | DOI:10.1371/journal.pone.0137717 September 14, 2015
1 / 22
Oral Microbiome Analysis of Denture Wearers
roles of these authors are articulated in the "author contributions" section. Competing Interests: GlaxoSmithKline (GSK) supported this project and were involved in the initial discussions at the conception of the study. David Bradshaw and Margaret Lambert are employed by GlaxoSmithKline and Gordon Ramage acted as a key opinion leader for GSK. There are no patents, products in development or marketed products to declare. This does not alter the authors' adherence to all the PLOS ONE policies on sharing data and materials, as detailed online in the guide for authors.
Conclusions This is the first study to provide a detailed understanding of the oral microbiome of denture wearers and has provided evidence that DS development is more complex than simply a candidal infection. Both fungal and bacterial kingdoms clearly play a role in defining the progression of DS, though we were unable to show a defined role for AMPs.
Introduction Healthcare improvements in the last century have led to an increasingly elderly population. Worldwide, 810 million people are aged 60 years or over, which is predicted to increase to at least two billion by 2050 (22% of the entire global population) [1]. In the EU alone the proportion of the population who are 65 years and older is predicted to reach 53% by the year 2025 [2]. This demographic change will result in significant challenges for oral healthcare delivery, to an increasingly aged population with declining oral health. As the population ages then oral diseases become more relevant with respect to their local and systemic impact, which can have profound implications for healthcare provision [3, 4]. The oral cavity is a complex environment that is continually exposed to numerous opportunistic microbial pathogens. These are kept in check by a robust arsenal of immune factors that maintain a healthy oral environment and prevent the development of disease. This arena has gradually become a key area of biomedical research, which has led to a greater understanding of the causes, pathogenesis and host response against oral disease, with the majority of research focussing on diseases affecting dentate individuals, such as gingivitis, periodontitis and caries. Conversely, there is very limited research regarding denture related disease. Despite major improvements in oral health worldwide, recent estimates report that the rate of edentulouness still varies from 7 to 69% of the worlds adult population [5, 6], and in the US and UK populations around one fifth wear some form of removable denture [7, 8]. This continued high prevalence should convince researchers that, there is a requirement to develop an understanding of the implications of dentures on oral and systemic health. The primary condition denture wearers suffer from is denture stomatitis (DS). This refers to inflammation of the oral mucosa and pathological changes associated with the wearing of dentures [9]. DS can be classified according to the severity of inflammation using a scale first described by Newton [10, 11]. The aetiology of DS is related to a variety of factors including poorly fitting dentures causing trauma and biological factors such as poor salivary flow, smoking or antibiotic treatment as well as microbial infection [12]. However, Candida albicans is generally attributed as being the main causative agent in DS affecting approximately 30–70% of denture wearers [9]. Candida spp. colonise the denture surface, forming co-aggregates with bacteria to build complex microbial communities known as biofilms. The majority of literature in this area focuses solely on Candida as the primary cause of infection, however, there is increasing evidence to suggest that this is very much a polymicrobial disease in which bacterial and fungal interactions play a role in disease pathogenesis [13, 14]. Several studies have isolated bacteria directly from the surface of dentures using standard microbial culturing techniques, primarily streptococci and staphylococci species [15–18]. However, these culture based methods are largely inadequate and unlikely to give a comprehensive representation of the polymicrobial population, which can contain up to 1011 microbes per milligram of denture plaque [19]. The advent of high throughput sequencing however has revolutionised our understanding of microbial ecosystems, and thus using this superior method we can for the first time gain an insight into the oral microbiome of denture wearers.
PLOS ONE | DOI:10.1371/journal.pone.0137717 September 14, 2015
2 / 22
Oral Microbiome Analysis of Denture Wearers
In addition to an incomplete understanding of the composition of denture plaque, we also have limited knowledge and understanding of the local host immune response. It has been established that the immune response is gradually impaired with increasing age [20, 21], but with the addition of loss of natural teeth an even greater rapid decline in host protective responses in the oral cavity is reported [22]. Antimicrobial peptides (AMP) including cathelicidin LL-37, histatins and defensins, which exhibit antimicrobial and immunoregulatory properties and protect mucosal surfaces against pathogens, have been found to be present in saliva. However, no studies have yet investigated salivary AMPs in denture wearers and the potential role they may play in the protection against denture plaque. Thus, a more in depth investigation is required into the host-microbiome relationship in the denture wearing individuals, which may help towards understanding the pathogenesis of DS. The aim of this study was therefore to carry out a detailed site specific analysis of the bacterial composition of the oral microbiome of denture wearers using high throughput 16S rRNA gene sequencing technology. Furthermore, to assess changes in the diversity and composition of the microbiome against candidal load to aid understanding of the potential bacterial-fungal interactions that may be important in DS. In addition, the relationship between host antimicrobial peptides and denture plaque composition was also investigated. This is the first study to provide such a detailed microbial analyses of denture biofilms.
Materials and Methods Study participants 123 denture wearing patients attending the University of Glasgow Dental School and Hospital were enrolled in the study. Convenience sampling was used based on the patient availability on recruitment days. Patients were recruited by a designated research nurse or a PhD student. Written informed consent was obtained from all participants. Ethical approval for the study was granted by the West of Scotland Research Ethics Service (12/WS/0121). All patients wore full or partial removable dentures. A team of qualified dental clinician were responsible for the collection of samples and the recording of clinical features. Data on patient age, gender, smoking status, routine oral hygiene regimens and any history of recent antimicrobial medication were recorded on a patient questionnaire. There was no age related exclusion criteria for this study. Newton’s classification method for DS was used to score the appearance of the patient’s palatal mucosa [11]. The following scores were applied; 0 = healthy mucosa, 1 = pin-point hyperaemic lesions (localized erythema), 2 = diffuse erythema (generalized simple inflammation), and 3 = hyperplastic granular surface (inflammatory papillary hyperplasia). For standardisation, all clinicians received training to calibrate scoring the extent of erythema. Patients were excluded from the study if they were pregnant, had previous radiotherapy for the treatment of head and neck malignancy, had been receiving antimicrobial/antifungal treatment, using prescription mouthwashes or had received immunosuppressant therapy within six months previous to sampling.
Clinical sample collection Clinical samples were collected as shown in Fig 1. Ethylene oxide sterilised swabs (Fisher Scientific, Loughborough, UK) were used to collect samples from the denture surface in contact with the palatal mucosa and the palatal mucosal surface covered by the denture. If any natural were teeth present the clinician took a supragingival plaque sample from a single tooth using a sterile dental probe, which was immediately placed into a 2 ml collection tube (Fisher Scientific) containing RNAlater1 (QIAgen, Manchester, UK). Dentures were removed from the patients’ mouth and placed in sterile bags (Fisher Scientific) filled with 50 ml PBS (Sigma-Aldrich,
PLOS ONE | DOI:10.1371/journal.pone.0137717 September 14, 2015
3 / 22
Oral Microbiome Analysis of Denture Wearers
Fig 1. Flow chart demonstrating the process of clinical sample collection and processing in the study. doi:10.1371/journal.pone.0137717.g001 PLOS ONE | DOI:10.1371/journal.pone.0137717 September 14, 2015
4 / 22
Oral Microbiome Analysis of Denture Wearers
Dorset, UK), then placed in a sonic bath (Ultrawave, Cardiff, UK) for 5 min to remove adherent denture plaque. The denture sonicate was then transferred to a 50 ml tube. Denture sonicate was centrifuged for 10 min at 3700 x g, and the plaque pellet re-suspended in 2 ml of RNAlater (QIAgen), as previously described [23]. Swab tips were removed and stored in RNAlater. Denture plaque, dental plaque and swab samples were all stored at -80°C. In total samples from 131 patients were collected, which included 131 denture swabs, 131 mucosal swabs and 79 dental plaque samples. During DNA extraction process not all samples had sufficient DNA to use for sequencing, therefore DNA from 108 denture samples, 87 mucosal samples and 63 dental samples remained for sequencing, collectively all these samples originated from 123 patients. Samples were further separated into distinct groups for direct comparison and analysis: health and DS groups, dentate and edentate groups and complete denture and partial denture groups.
Collection of saliva Samples of whole unstimulated saliva were obtained by expectorating into a SaivaBio collection tube (Salimetrics, Suffolk, UK). Patients were given a maximum of 5 min to provide a sample. Saliva was clarified by centrifugation at 10,000 g for 10 min and then stored at -80°C.
Candida isolation One ml of denture sonicate was used to prepare ten-fold serial dilutions ranging from 100– 10−3. The dilutions were then used to perform colony forming unit (CFU) counts using Colorex Candida plates on which C. albicans colonies appeared green, C. glabrata were pink and C. tropicalis were blue. (E&O Laboratories, Bonnybridge, UK). One hundred μl of each serial dilution was spread across each plate, and then incubated at 30°C for 48 hrs. CFUs forming on plates were then counted and the average number of Candida cells colonising each denture was calculated.
DNA isolation Dental plaque samples were centrifuged for 15 min @ 13000 x g the supernatant was removed and the sample resuspended in 150 μl TE buffer. Swab samples were sonicated for 30 sec, and then the sonicated fluid was transferred into a deep well plate and centrifuged for 15 min at 13000 x g. The precipitate was then resuspended in 150 μl TE buffer. All samples were then transferred to a plate with each well containing 0.25 ml of lysis buffer (AGOWA mag Mini DNA Isolation Kit, AGOWA, Berlin, Germany), 0.3 g zirconium beads (diameter, 0.1 mm; Biospec Products, Bartlesville, OK, USA) and 0.2 ml phenol. The samples were homogenized with a Mini-beadbeater (Biospec Products) for 2 min. DNA was extracted with the AGOWA mag Mini DNA Isolation Kit.
Quantitative PCR Real time qPCR was performed to determine the concentration of bacterial DNA per sample as the original DNA extracted may contain DNA from other microbes such as fungi. Primers and a probe for the 16S rRNA gene were used, (F: TCCTACGGGAGGCAGCAGT, R: GGACTAC CAGGGTATCTAATCCTGTT Probe: 6FAM-CGTATTACCGCGGCTGCTGGCAC-BBQ). As an internal control for PCR inhibition a qPCR of PhHV (Phocid herpesvirus type 1 gB gene) was also performed as described by Watzinger et al (2004) [24]. The total reaction volume was 20 μl, including 3 μl of DNA. Reactions contained a 26PCR Probe Master Mix (Roche), for 16S rRNA, 7.5 pmol primers and 3.8 pmol probe. For PhHV 1.8 pmol primers
PLOS ONE | DOI:10.1371/journal.pone.0137717 September 14, 2015
5 / 22
Oral Microbiome Analysis of Denture Wearers
and 0.4 pmol probe of each primer was used. qPCR was carried out using the Light cycler LC480-II (Roche Diagnostics, Switzerland) under the following conditions: an activation step of 10 min at 95°C, followed by 50 cycles consisting of a denaturation step at 95°C for 30 sec, an annealing step at 60°C for 30 sec, and an extension step at 72°C for 30 sec. Bacterial 16S rDNA concentrations (CFU/ml) were determined from standard curves of E. coli K12 cultures.
PCR amplification and Illumina sequencing Amplicon libraries of the V4 hypervariable region of the 16S rRNA gene were generated for each of the individual samples. PCR was performed using the forward primer 515F (GTGCCA GCMGCCGCGGTAA) and the reverse primer 806R (GGACTACHVGGGTWTCTAAT) [25]. The primers included Illumina adapters and a unique 8-nt sample index sequence key [26]. The amplification mix contained 2 units of Phusion HotStart II High fidelity polymerase (Thermoscientific), 1 unit Buffer Phusion HS II [5x], including 1.5 mM MgCl2 (Thermoscientific), 0.2 mM dNTP (Thermoscientific, Germany) and 1 μM of each primer. To each reaction 1 ng of DNA template was added. After denaturation (98°C; 30 sec), 35 cycles of denaturation (98°C; 10 sec), annealing (55°C; 30 sec), and extension (72°C; 30 sec) were performed. Individual amplicon libraries were analyzed for DNA content with the fluorescent Quant-iT™ PicoGreen1 dsDNA Assay Kit (Invitrogen). The libraries were pooled in equimolar amounts. The amplicons were purified by means of the IllustraTM GFXTM PCR DNA and Gel Band Purification Kit (GE Healthcare, Eindhoven, the Netherlands). The quality and the size of the amplicons were analyzed on the Agilent 2100 (Santa Clara, CA, USA). The amplicon was sequenced in paired end mode on a MiSeq sequencing system (Illumina, Eindhoven, the Netherlands) with the v2 kit (Illumina) [26, 27].
Sequencing data analysis Reads were first quality filtered using Trimmomatic v0.32, [28]. Next, the reads were merged using fastq-join implemented in QIIME v.1.8.0 [29]. Sequences were clustered into operational taxonomic units (OTUs) using USEARCH v7.01090 [30], after quality filtered with usearch (maxee 0.5).The representative sequence of each cluster was assigned a taxonomy using the RDP classifier [31]. The Ribosomal Database Project: improved alignments and new tools for rRNA analysis. Nucl Acids Res 37: D141–145. doi: 10.1093/nar/gkn879) (QIIME v.1.8.0) (Greengenes v13.8 97_otus set) with a minimum confidence of 0.8.
ELISAs A range of antimicrobial peptides which are commonly associated with the oral cavity were selected for assessment in our patient’s saliva [32–34]. The following AMPs were assessed by ELISA, LL-37, Calprotectin, Lactoferrin, HNP1-3 (Hycult biotech, The Netherlands), Histatin 5 (Stratech scientific, Suffolk, UK) and Beta defensin 1 (BD1) (Peprotech, London, UK) as per manufacturer’s instructions. Clarified saliva samples were diluted 1:5 in assay buffer (PBS, 0.5% BSA, 0.1% Tween20). Results were calculated using a 4-parameter curve fit, quantifying colorimetric changes at 630 nm (BMG-Labtech, Ortenberg, Germany).
Statistical analyses The data set was randomly sub-sampled to 770 reads per sample (minimum number of reads per sample was 776) to include the maximum number of samples for analysis. OTU datasets were reduced by log2 transformation so as to carry out principal component analysis (PCA)
PLOS ONE | DOI:10.1371/journal.pone.0137717 September 14, 2015
6 / 22
Oral Microbiome Analysis of Denture Wearers
and diversity statistics (Shannon diversity index and Dominance index); the analysis was carried out using PAST software [35]. A one-way ANOVA test was applied to compare diversity statistics at oral microbiome sites using GraphPad Prism software (version 4; La Jolla, CA, USA). PCA was used to reduce the dimensionality of the OTU dataset. 258 OTU’s were entered into the PCA. A scree plot was used to determine how many components emerged. Factor loadings above 0.15 on a component were considered to have a strong association with that component and were deemed to be the most informative in describing the microbiome components. To determine if distinct clusters formed for each group on the PCA plots, new variables were created for each principle component by using the factor loadings as regression coefficients, producing a score for each sample. These scores were then used as outcome variables to compare between groups (using t-tests where appropriate- dentate/edentate and complete/partial dentures groups). The contribution of each bacterial class was calculated in terms of proportion to the overall sample, percentages were log transformed and a t-test was used to compare health and DS groups. Diversity statistics were compared via a t-test with GraphPad Prism v5. Spearman’s rank correlation was used to assess correlations between the abundance of individual bacterial classes or genera with the proportion of Candida found on dentures (CFU counts), using SPSS version 20. Salivary concentrations of AMPs were compared between healthy and DS groups, dentate/ edentate groups and complete/partial groups (S1 Table). Data were log transformed and analysed using a t-test with prism. Furthermore Spearman’s rank correlation was used to assess correlations between the abundance of individual bacterial classes or genera with the concentrations of salivary AMPs found on dentures. To visualize the relationships and associations of the microbiomes with environmental variables canonical correspondence analysis (CCA) was applied. This form of analysis, carried out using PAST, allows the visualisation of OTU distribution and sample group distribution in relation to a number of environmental variables. Environmental variables included were Candida CFU counts and salivary concentration of a number of AMPs. The significance of each of the CCA axes was calculated by permuting the data 999 times. Candida CFU counts were compared between healthy and DS groups, dentate/edentate groups and complete/partial groups. Data were log transformed and analysed using a t-test with GraphPad Prism v5.
Study design The study was designed as a pilot study and was initially only powered to detect a biologically meaningful association between diseased and healthy mouths and microbiome composition, and therefore was not originally powered to detect differences between additional variables including, denture type, dentate status, Candida levels and salivary AMP levels. Thus, non-significant results between these variables are not necessarily absence of effect, but a result of not achieving the full sample size required. S2 Table provides a list of the statistics carried for all the analysis. Dentate and edentate groups were compared separately from complete and partial denture groups, as having a complete denture only applied to the upper denture from which the sample was taken. The patient however may have had natural teeth remaining on the mandible and therefore could not be classed as edentulous. The inter-correlation between complete and edentulous or partial and dentate groups are shown in Table 1. Thus given the variation in group numbers, analysis was carried out on both datasets.
PLOS ONE | DOI:10.1371/journal.pone.0137717 September 14, 2015
7 / 22
Oral Microbiome Analysis of Denture Wearers
Table 1. Patient demographics.
Patients n = 123
Gender n (%)
Mean age
Type of Denture n (%)
Dentate status
Disease status n (%)
Mean age of denture
(Male) (Female)
(years)
(Complete upper) (Partial upper)
(Edentulous) (Dentate)
(Healthy) (Disease)
(years)
43 (35)
70.7 ± 11.5 (min: 33, max: 95)
4.6 ± 5.3 (min:0.2, max: 40)
80 (65)
82 (67)
56 (46)
78 (63)
41 (33)
67 (54)
45 (37)
doi:10.1371/journal.pone.0137717.t001
Results Patient demographics Of the 123 patients that participated in this study, 43 were male and 80 were female (Table 1). The mean (SD) patient age was 70.7 (11.5) years (min: 33, max: 95) with an average (SD) denture age of 4.6 (5.3) years (min: 0.2 max: 40). Clinical diagnoses indicated that 63% of participants had a healthy oral mucosa and the remaining 37% had an inflamed mucosa of varying severity, and were suffering from DS. The majority of patients (67%) wore complete upper dentures, while the remainder (33%) had a partial upper denture with 1 natural teeth remaining.
Illumina sequencing output Across all of the samples 632 OTUs were identified, with an average of 94 OTUs per sample (SD 39; min 25, max 254), 536 of which contained a minimum of 5 reads. The data was subsampled to 770 reads per sample in order to avoid bias of variable sample size. After sub-sampling 502 OTUs remained with an average of 46 OTUs per sample (SD 18; min 11, max 121). The samples were categorised into five main phyla which represented 99.6% of the reads: Firmicutes (40.6%), Actinobacteria (23%), Bacteriodetes (22.2%), Proteobacteria (9.8%) and Fusobacteria (4%).
Oral microbiome by sample site Samples were categorised into groups according to sample site: denture (n = 108), mucosal (n = 87) and dental (n = 63), with 337, 414 and 306 OTUs identified, respectively. On average, denture samples had 42 OTUs (SD 14; min 13, max 87), mucosal 49 OTUs (SD 20; min 11, max 121) and dental 51 OTUs (SD 37; min 16, max 306). Interestingly, denture plaque samples were found to have significantly less OTUs compared to dental, (p
